Itacism: Flexibility between ι, η, ει in the second century

Covering all topics of history and the interpretation of texts, posts here should conform to the norms of academic discussion: respectful and with a tight focus on the subject matter.

Moderator: andrewcriddle

Post Reply
User avatar
Peter Kirby
Site Admin
Posts: 8483
Joined: Fri Oct 04, 2013 2:13 pm
Location: Santa Clara
Contact:

Itacism: Flexibility between ι, η, ει in the second century

Post by Peter Kirby »

I recently came across a linguistics thesis that summarizes, among other things, how linguists view the "evolution of the grapheme η" (starting page 15), primarily following Threatte's 1980 account. Starting on page 21 is a discussion of the early Roman period (27 BCE to 284 CE).

The Shape of Eta: Evolutionary Phonology and the Development of Attic Greek by Briana Grenert
https://scholarship.tricolib.brynmawr.e ... 7b/content

They provide this table for a "history" of eta, with reference to Attic Greek (not necessarily koine Greek):
history-of-eta.jpg
history-of-eta.jpg (449.39 KiB) Viewed 631 times

After 150 CE, cases of confusion between η and ι become more frequent - first in crude texts, and then in other texts as well. Threatte proposes that the pronunciation of η as /i/ was probably substandard. From the fact that there is still confusion between η and ε at the same time as between η and ι, it is likely there were multiple pronunciations available. This overlap is present both in inscriptions and papyri, and persists until the Late Roman Period. ...

In the second century, there is a resurgence of instances of /i/ represented as ι, η, ει. Table 10 gives three attested spellings of the male given name Charisios.

charisios.jpg
charisios.jpg (52.17 KiB) Viewed 631 times

This flexibility between ι, η, or ει is especially in papyri (less so in monuments). This spelling does not resurge on monuments until later in the Roman period. ...

In the monuments of the poor, confusion between η ~ ι begin to appear again. Even in the late Roman period, there is still some confusion between η ~ ε, so it seem that η was probably pronounced /e/ by some. See Table 12. ...

The wealthy seem to hold out with η as either /ε:/ or /e/ throughout the fourth century; however the battle was lost and by the end of the fourth century, when η was pronounced more exclusively as /i/.

interchange.jpg
interchange.jpg (329.12 KiB) Viewed 631 times
The thesis writer (early in their academic career...) draws primarily on Threatte, whose book on "Attic inscriptions" has (naturally) a focus on Attic inscriptions found in Greece. Grenert is still able to find their examples from Threatte, who also has a few more in his book: https://archive.org/details/grammarofat ... 8/mode/2up (page 169)

πέψης (another spelling of πέψις)
https://inscriptions.packhum.org/text/235357?hs=195-201

Δηραδιώτης (another spelling of Δειραδιωτης)
https://inscriptions.packhum.org/text/8298?hs=68-80

καταβαλῆτε (another spelling of καταβαλεῖται / καταβαλεῖτε)
https://inscriptions.packhum.org/text/15713?hs=167-178

ἠσελεύσοντε (another spelling of εἰσελεύσονται / εἰσελεύσοντε)
https://inscriptions.packhum.org/text/345530?hs=66-79

Gignac, who focuses on koine Greek, and who makes wide use of non-literary papyri, describes the interchange of graphemes in Egypt in the early Roman period this way (page 235 of https://archive.org/details/gignac-a-gr ... ology-1975 - A Grammar Of The Greek Papyri Of The Roman And Byzantine Periods, Vol. 1 Phonology):

The process of itacism, which resulted in the eventual identification of the sounds originally represented by ι, ει, η, ηι, οι, υ, υι in /i/, was well advanced in Egypt by the beginning of the Roman period. ει and ι are alternate representations of /i/; η and ηι are identified; οι, υ, and υι all represent /y/. Moreover, there is a very frequent interchange of η with ι and ει, indicating that η also represented /i/ at least in the speech of many writers. On the other hand, there is a frequent interchange of η with ε (and sometimes with its phonetic equivalent αι) throughout the Roman and Byzantine periods, in similar documents and sometimes in identical phonetic conditions and even in the same words in which an interchange of η with ι or ει is found. There is also an occasional interchange of ε (αι) with ι and ει. ...

η x ι.
This interchange occurs very frequently in all phonetic conditions throughout the Roman and Byzantine periods.

Gignac provides numerous examples in both directions. I've provided four screenshots with some of them in one of those directions, i.e., where the spelling uses eta (instead of iota or epsilon-iota). There are roughly an equal number of examples in the other direction in the book.

gignac-1.jpg
gignac-1.jpg (787.9 KiB) Viewed 631 times
gignac-2.jpg
gignac-2.jpg (568.89 KiB) Viewed 631 times
gignac-3.jpg
gignac-3.jpg (825.85 KiB) Viewed 631 times
gignac-4.jpg
gignac-4.jpg (824.25 KiB) Viewed 631 times

Gignac's conclusion on p. 241 is that the sounds represented by the graphemes had merged as /i/ in general usage in Egypt by the second century. This conclusion is regarding Koine Greek, especially in Egypt, not regarding Attic Greek, which was Threatte's concern. This explains the differences between the two, with Threatte and likewise Grenert emphasizing the limited scope of the interchange before the mid-second century in Attic Greek. Synthesizing the conclusions of Threatte and Gignac, Attic Greek in the second, third, and fourth century was different than the Greek Koine (in Egypt, at least) with respect to the phonology of these graphemes.

This interchange of with η and ι reflects the phonological development of the Greek Koine, in which the sound originally represented by η generally merged with /i/ by the second century A.D.

Based on this linguistic data, we can reach the conclusion that the interchange of the graphemes η and ι in the second century wasn't unidirectional. Threatte and Gignac both provide examples of spelling changes with substitution of graphemes in both directions. There's no pattern suggesting that one was more likely than the other or presented only in a more limited set of circumstances.

The interchange occurs very frequently when the phonology is the same in general use (as Gignac concludes that it was in the second century koine Greek of Egypt). It also occurs in situations where the phonology is the same only in substandard use (as suggested by Threatte regarding second century Attic Greek). At the same time, but in different places and perhaps different social circles (as described above, especially with the contrast of speakers in Egypt instead of Greece), some second century Greek speakers identified the sounds represented by the graphemes iota and eta, and other second century Greek speakers differentiated them.

So this is where I tie this back a bit into this forum (but without involving any arguments specifically about Christian texts). Martijn Linssen has expressed an understanding about the interchange of graphemes that is misguided (but commonly expressed online):
mlinssen wrote: Wed Oct 11, 2023 3:47 amitacism involves (we’ll come to that in Itacism/iotacism) the exact opposite, namely an ‘e’ or eta changing into an ‘i’ - and not vice versa;

Linssen, Martijn. Gospels, Epistles, Old Testament: The order of books according to Jesus Chri st (p. 84). Kindle Edition.
mlinssen wrote: Wed Oct 11, 2023 3:47 amitacism describes the movement towards the iota, and not away from it:

Linssen, Martijn. Gospels, Epistles, Old Testament: The order of books according to Jesus Chri st (p. 207). Kindle Edition.
mlinssen wrote: Wed Oct 11, 2023 3:47 amApologists stick to phrases as used by e.g. the Brill editors as demonstrated: ‘the spelling χρηστός may be used erroneously for χριστός, the result of itacism.’ Itacism entails the exact opposite of what is claimed here. Itacism gets abused by Christian apologists to iron out the countless wrinkles in the textual history, and a plausible explanation for the complete absence of any explanation whenever they wield their blunt argument is that they intentionally try to disinform and mislead the reader, so that they can falsify the evidence without questions asked - the alternative option is ignorance and incompetence.

Linssen, Martijn. Gospels, Epistles, Old Testament: The order of books according to Jesus Chri st (pp. 204-205). Kindle Edition.
Martijn has several arguments that "everything in that textual tradition amply testifies to itacism not being applicable," and I'm not actually arguing against that right now. I'm not arguing about any particular instances, and I'm not trying to argue about the Christian texts. In this post I'd like to stay focused on the general topic of "itacism" in Greek instead of the particular subjects of when and where it might be applicable in the Christian textual tradition. Martijn could be partly or mostly right about what he focuses on, even while misunderstanding the general concepts here. One thing at a time.

Martijn argues, among other things, that words such as χρηστός and χριστός would not have been pronounced the same way. He argues that this would create misunderstandings, especially because they are both used as an adjective, so they must have been pronounced differently.
mlinssen wrote: Wed Oct 11, 2023 3:47 amEven arguing for χρηστός being pronounced like χριστός simply doesn’t hold because it would equate these two very different words to one another. It is nigh impossible to mistake one word for another, that is the whole concept of words and languages - if that weren’t the case, every language could consist of one single word alone. Why are homonyms and homophones such pesky things? Precisely because they appear to be identical yet aren’t - and it is only then and there, when and where, confusion arises, exactly because content alone doesn’t suffice any more, and needs support from context. In English many homophones exist, words that are spelled differently yet sound the same: ‘heal’ vs ‘heel’, ‘beat’ vs ‘beet’ and ‘real’ vs ‘reel’ are some examples, and very often these words are from different grammar classes, e.g. a verb and a noun. Yet χρηστός and χριστός both are used as adjective, which certainly would create complete misunderstandings. Just picture that e.g. ‘holy’ and ‘hollow’ are pronounced exactly the same, perhaps ‘hollyw’ - imagine the utter confusion?

Linssen, Martijn. Gospels, Epistles, Old Testament: The order of books according to Jesus Chri st (p. 209). Kindle Edition.
Entirely removed from these theological debates, Gignac provides a set of empirical data in support of the conclusion that in general use in the second century koine Greek of Egypt, the graphemes η and ι both represented the same sound, /i/. The same is true of Byzantine Greek, the widespread form of Greek by the fifth century. Threatte's account of Attic Greek completes the outline of the picture, showing that the distinct pronunciation was still being preserved in the Attic Greek of the second to fourth century, albeit with interference from the contemporary "substandard" pronunciation that pronounced them both as /i/.

Either Martijn is right or the linguists studying ancient Greek are. If we followed Martijn's line of reasoning, the pronunciation of Gignac's second century koine Greek speakers in Egypt and the pronunciation of Byzantine Greek speakers was not possible. Linguists would be able to conclude their research tidily and say that both graphemes never represented the sound /i/ so long as the words that Martijn mentions (χρηστός and χριστός) were in the language. However, Martijn's argument is based on an unfounded confidence in his powers of deductive reasoning, while the linguists draw on a wide variety of empirical data, developing a linguistic history of Greek that leads to the very "itacism" that was revealed with evidence, centuries later, by scholars like Erasmus (to a public unfamiliar with evidence that the sounds of the two letters were different in Attic Greek because they were the same in contemporary Greek).

I need to mention again what conclusions I am actually stating right now because the whole thing is a hornet's nest. I'm trying to give the reader just a few pieces of knowledge here - the honey, if you will, without the stinging of a thousand arguments. You can take that knowledge and combine it with the bread and butter of any theory of your preference. I'm not knocking your favorite theory on these Christian theological terms. I'm not even saying that "itacism" was ever a factor (and I'm not saying that it never was either, for that matter). Again, one thing at a time. We'll be much better prepared if we try to understand the Greek before we try to apply what we think about the Greek to the particulars of a very contested subject.

Those conclusions are, once again, this: the interchange of the graphemes η and ι in the second century (and later) wasn't unidirectional. Threatte and Gignac both provide examples of spelling changes with substitution of graphemes in both directions. There's no pattern suggesting that one was more likely than the other or presented only in a more limited set of circumstances. Gignac provides a set of empirical data in support of the conclusion that in general use in the second century koine Greek of Egypt, the graphemes η and ι both represented the same sound, /i/. The same is true of Byzantine Greek, the widespread form of Greek by the fifth century. Threatte's account of Attic Greek completes the outline of the picture, showing that the distinct pronunciation was still being preserved in the Attic Greek of the second to fourth century, albeit with interference from the contemporary "substandard" pronunciation that pronounced them both as /i/.

Martijn presents several arguments about the particulars of the Christian texts and the two words that are his focus. Perhaps some, most, or even all of them are right. That's not the point here.

However, in the course of doing so, he relies on a false understanding of the phonology / phonologies of Greek in the ancient world. Without interacting with the work of any linguists who have written about ancient Greek, he argues based only on his personal, homespun disbelief against the conclusion of those linguists. The conclusion of the linguists is that there were ancient Greek speakers who were pronouncing the graphemes η and ι both as /i/ as explained above. He misunderstands the phenomenon of the interchange of graphemes when it comes to interchange due to "itacism," given that he maintains that it operated in one direction: "namely an ‘e’ or eta changing into an ‘i’ - and not vice versa." His general understanding of this subject is impoverished and incorrect.

This post doesn't present anything regarding Coptic, which is a different language and requires a different post. I hope you enjoyed my Ted talk, thank you.
Secret Alias
Posts: 18748
Joined: Sun Apr 19, 2015 8:47 am

Re: Flexibility between ι, η, ει in the second century

Post by Secret Alias »

Love your post. Clement of Alexandria was an Atticist and clearly distinguished chrestos from having anything to do with anointing. Here is a practical benefit to acknowledging the pre-existing Alexandrian tradition of Mark. With Mark, Christianity rises from the gutter. Hebrews was not written in koine. Clement wrote in better Greek than the LXX.
davidmartin
Posts: 1610
Joined: Fri Jul 12, 2019 2:51 pm

Re: Flexibility between ι, η, ει in the second century

Post by davidmartin »

It seems to me the question on itacism only relates to a sub-set of the argument ML makes since his argument certainly isn't based solely on itacism
it would be good to clarify which of the book's points (chapters/page numbers) apply here
i mean, otherwise the impression might be it undermines his whole thesis for those that have not read the book, so this is a narrow focus on that one specific area?
when you say "his argument" it relates just to his argument as it relates to itacism...
User avatar
Ken Olson
Posts: 1341
Joined: Fri May 09, 2014 9:26 am

Re: Flexibility between ι, η, ει in the second century

Post by Ken Olson »

davidmartin wrote: Sat Jan 20, 2024 3:26 pm It seems to me the question on itacism only relates to a sub-set of the argument ML makes since his argument certainly isn't based solely on itacism
it would be good to clarify which of the book's points (chapters/page numbers) apply here
i mean, otherwise the impression might be it undermines his whole thesis for those that have not read the book, so this is a narrow focus on that one specific area?
when you say "his argument" it relates just to his argument as it relates to itacism...
I thought Peter Kirby was perfectly clear on what the point he was criticizing Martijn Linssen on was, your concern about how people who have not read [Linssen's] book might take it notwithstanding. Kirby never uses the two words 'his argument' together, he identifies the arguments to which he is responding, and the one time he uses 'argument' in the singular, it's in the middle of a paragraph and an attentive reader should be able to grasp from the preceding sentences that it concerns Linssen's argument about the pronunciation of χρηστός and χριστός (which was given in a block quotation earlier).

I am wondering if anyone has any criticisms concerning the substance of Peter Kirby's argument about itacisms - that is, with its factual or logical content. It would not surprise me if no one had, as itacism is a very specialized area outside of most people's area of knowledge (me included).

The criticisms we've had so far look like tone policing.

Best,

Ken
User avatar
Peter Kirby
Site Admin
Posts: 8483
Joined: Fri Oct 04, 2013 2:13 pm
Location: Santa Clara
Contact:

Re: Flexibility between ι, η, ει in the second century

Post by Peter Kirby »

davidmartin wrote: Sat Jan 20, 2024 3:26 pm it would be good to clarify which of the book's points (chapters/page numbers) apply here
i mean, otherwise the impression might be it undermines his whole thesis for those that have not read the book, so this is a narrow focus on that one specific area?
I was very clear to say things that show I don't think it would "undermine his whole thesis." Please review the OP if you're in doubt. There were also page numbers.

I spent three to four paragraphs anticipating that kind of thing just so that people wouldn't think / reply this way...
User avatar
Peter Kirby
Site Admin
Posts: 8483
Joined: Fri Oct 04, 2013 2:13 pm
Location: Santa Clara
Contact:

Re: Flexibility between ι, η, ει in the second century

Post by Peter Kirby »

Regarding the phonology of Coptic, the information is not as easily accessible as it is for Greek. I went through a lot of what's on https://copticsounds.wordpress.com/resources/ to try to find some answers. It's not easy reading.

The account that made the most sense to me (being consistent with the data and clearly outlined) was that of James Allen's Coptic: A Grammar of Its Six Major Dialects (2020), in particular his remarks on "Greek words."

allen-1.jpg
allen-1.jpg (916.69 KiB) Viewed 323 times
allen-2.jpg
allen-2.jpg (779.54 KiB) Viewed 323 times
allen-3.jpg
allen-3.jpg (374.49 KiB) Viewed 323 times

This series of articles by Girgis is significant for surveying Greek loan words and noting examples of interchange in Coptic (with these first two in that four-part series of particular interest). The screenshots show some examples of grapheme interchange that resemble the same patterns found in koine Greek as noted in the OP.

https://copticsounds.files.wordpress.co ... bsac17.pdf
https://copticsounds.files.wordpress.co ... bsac18.pdf
https://copticsounds.files.wordpress.co ... bsac19.pdf
https://copticsounds.files.wordpress.co ... bsac20.pdf

oi-to-eta.jpg
oi-to-eta.jpg (1.45 MiB) Viewed 323 times
eta-to-iota.jpg
eta-to-iota.jpg (1.3 MiB) Viewed 323 times
coptic-eta-iota.jpg
coptic-eta-iota.jpg (1.27 MiB) Viewed 323 times

The examples support Allen's remarks on Greek words in Coptic. Layton notes the same phenomenon (A Coptic Grammar, p. 33) regarding the interchange of graphemes in Greek words used in Coptic.

Like Allen (above), Gignac in "Influence of Coptic on the Phonology of the Greek of Egypt" says that Coptic ⲏ was "bivalent," which is a technical term that means (as defined by Woolard in "Simultaneity and bivalency as strategies in bilingualism") "simultaneous membership of an element in more than one linguistic system'' (in this case, Koine Greek and Sahidic Coptic). This is a reference to the way that bilingual speakers are aware of the use of the element in both linguistic systems when they communicate, allowing them to draw on one or the other (or both).

Allen wrote above that the most likely explanation is that the phoneme represented by ⲏ was /i/ when the grapheme was used as an element borrowed from the Greek linguistic system and that it was /ε/ or perhaps /e/ when used as an element found in the Egyptian linguistic system. Allen's suggestion explains the data, such as the examples provided by Girgis, and makes sense in terms of the development of Coptic:

(1) There were previous attempts at representing the Egyptian language in Greek script, and they were modeled (partly) on the phonetic values of classical Greek as taught by the grammarians. So it makes sense that the transcription of native Coptic words would be made phonetically based on some of these existing conventions. Prior to the development of Coptic writing, only native Egyptian words were represented in this script.

(2) At the same time, koine Greek was already dominant in Egypt before the development of Sahidic Coptic writing in the third century, so it makes sense that the Greek loan words (many of which had already become part of Coptic as spoken) did not acquire a different pronunciation in Coptic just because of their spelling. But this meant the grapheme ⲏ would become bivalent in Coptic; depending on the context, it could represent one of two different phonemes.

The short of it is that the writing of native Coptic words would not often involve the interchange of ⲏ and ⲓ because they would represent different phonemes in Coptic. Since the idea of different phonemes in linguistics is that the speakers would recognize them as meaningfully different, capable of indicating the use of different words, those sounding out the native Coptic words could easily tell the difference by ear.

On the other hand, when encountering words that were borrowed from koine Greek, most commonly in the same pronunciation as koine Greek itself, then both ⲏ and ⲓ would represent the same phoneme, /i/. This would allow for an understandable interchange of ⲏ and ⲓ in these words, as we find in some of the examples above.
User avatar
Peter Kirby
Site Admin
Posts: 8483
Joined: Fri Oct 04, 2013 2:13 pm
Location: Santa Clara
Contact:

Re: Flexibility between ι, η, ει in the second century

Post by Peter Kirby »

Ken Olson wrote: Sat Jan 20, 2024 4:30 pm itacism is a very specialized area outside of most people's area of knowledge
I enjoyed learning about it and sharing what I found here!
Post Reply